A Finer-Grained Motivational Model of Behaviour Sequencing

In this paper we consider three points. 1) Etho-logical behaviour models are not grounded in exhibited behaviour-the models have not been developed far enough, indeed, work in the ani-mat community may help ethology; 2) In order to assess agent performance we should try to develop a functional account of behaviour whilst developing a mechanistic account; 3) Planning and reasoning is not necessary to exhibit sophisticated appetitive behaviour sequences. To illustrate these points, our model of behaviour selection in a self-suucient autonomous agent is based on an ethological motivation model which extends traditional ethological modelling in the direction of implementable behaviours. The model is rstly justiied from a functional approach and then tested and shown to have improved performance in a two-dimensional continuous simulated environment. It is noted that the model exhibits what could be described as planning behaviour.